3D Object Geometry from Single Image
The past few years have witnessed remarkable advancements in 2D image understanding driven by deep learning. But when it comes to the scenarios involving interactions between human, robot and the world, we need to extract not only 2D visual information but also 3D geometry of the scene from images. This talk focuses on our recent progresses on recovering 3D object geometry, such as 3D structure and pose of rigid and articulated objects, from monocular imagery by levering data-driven representations learned from both 2D and 3D data. In particular, I will first introduce an approach to 3D human pose reconstruction based on the sparse representation of 3D human poses and CNN-based 2D pose predictions. I will discuss how to jointly optimize structural and viewpoint parameters with convex programming and how to account for the uncertaintiesin 2D pose predictions with an EM algorithm. Next, I will show that, using semantic keypoints and CAD models, it is feasible to estimate the 6-DoFpose of a rigid object from a cluttered image with a precision allowing a robot to grasp the object. Finally, I will introduce how to build object-class models from images of different instances with a multi-image matching algorithm that optimizes the cycle consistency of feature correspondences. n
Xiaowei Zhou is a Postdoctoral Researcher in the Computer and Information Science Department at University of Pennsylvania. His research interests are on computer vision and robotics with a focus on object pose estimation, shape reconstruction, human pose estimation and data association.His current work attempts to combine 3D geometry, deep learning, optimization and statistical approaches to extract both semantic and geometric information of 3D scene from visual data. Xiaowei Zhou obtained his Bachelor’sdegree in Optical Engineering from Zhejiang University, 2008, and his PhD degree in Electronic and Computer Engineering from The Hong Kong University of Science and Technology, 2013.